In one sentence
Optimize total architecture cost by matching capacity, pricing, storage lifecycle, and data-transfer design to business requirements.
Key points
- Right-size resources and align capacity with demand using Auto Scaling, serverless, schedules, and managed services.
- Use On-Demand, commitment discounts, or Spot according to workload stability and interruption tolerance.
- Apply storage classes and lifecycle rules while checking retrieval time, access frequency, and deletion constraints.
Exam takeaway
Track tags, budgets, metrics, NAT and transfer charges, and never trade away required security, reliability, or performance.
Compute, database, and network additions
- Right-size EC2, use commitments for a stable baseline, Spot for interruptible work, and Auto Scaling for variable demand.
- Evaluate RDS size, storage, replicas, backups, and cache together; Read Replica, Multi-AZ, and ElastiCache solve different problems.
- Use Cost Explorer to locate bill changes, Budgets for thresholds, and Pricing Calculator before an architecture change.
- Compute Optimizer recommends sizes and Trusted Advisor checks broader practices; validate every recommendation against peaks and reliability.
- Data-transfer cost depends on direction, Region, AZ, and path. S3 and DynamoDB gateway endpoints have no hourly or data-processing fee, while interface endpoints do.
Pricing principles
Pay as you go, save when you commit, and pay less by using more. The lowest unit price is not necessarily the lowest total cost once requests, retrieval, transfer, operations, availability, and recovery are included.